ChatGPT rated "unacceptable risk" for teens as parental alerts failed on 4,000+ test prompts
The Decoder · rss · 2026-10-07
Common Sense Media's Youth AI Safety Institute audited ChatGPT's teen safety features with over 4,000 test prompts and rated the service an "unacceptable risk" for minors.
Conversations about suicide and self-harm on test accounts never triggered a parental alert, meaning OpenAI's parental notification mechanism failed in practice. The institute is calling for teens to be locked out of the service until its safety can be independently verified.
More from Safety
- Mumsnet AI Prompt Leak Sparks Trust Crisis Over Hidden AI Content — nordicinst · 2026-10-07
- Finland orders Google to halt work on two data centers over environmental reviews — KyeGomezB · 2026-10-07
- Hidden Chain-of-Thought Extracted from GPT-6 and Claude via Standard API Tool Calling — jiqizhixin · 2026-10-07
- Warden Infostealer Hunts AI Agent Keys, Exfiltrates Claude CLI Tokens — Malwarebeasts · 2026-10-07
- Google Opens SynthID Detector to Everyone, Detects AI Content from OpenAI, NVIDIA and More — GoogleDeepMind · 2026-10-07
- Google launches SynthID.com, its AI watermark detector now covers all partner models — Ars Technica AI · 2026-10-07